How's the AI

[ChuckBerger]

This game looks interesting, and I'm willing to overlook a lot in strategic WW2 wargames. But AI is a game-maker, or game-breaker for me.

The one thing about the Wargamer review that really gave me pause is the review of the game's AI. The summary is that it does OK once units are lined-up face to face on a battle from, but that the strategic AI is basically incompetent.

I mostly play single-player these days, and don't have much interest in buying a game with an AI that won't provide a plausible opponent as a whole, so I'm inclined to give WarPlan a miss.

But thought I'd check in with devs and users here first, before I write it off entirely.

[gwgardner]

I find the AI competent on the higher difficulty levels. I've seen comments here from players who've completed the 1939 Campaign scenario, on normal difficulty, and they say the AI gave them a fight. I've also seen AARs where the human player whipped up on the AI pretty badly, on the highest difficulty level.

On higher difficulty levels the AI gets experience and supply advantages, but otherwise plays by the same rules. That supply issue is what makes the game tick. It's related to unit 'effectiveness.' Higher effectiveness encourages the AI to be bolder and more stalwart. Lower effectiveness tends to make it less willing to take risks. A simple example from my tests: on normal difficulty, it can take the AI 5 or 6 turns to take Poland; on highest difficulty, the AI will take Poland in two turns.

I suspect there will be changes/improvements to the AI on a continuing basis. The developer seems dedicated to the long-term.

[tyronec]

I have played the game twice as Axis against the AI. The first time on moderate difficulty settings and was winning too easily so abandoned the game. The second time on hardest difficulty; it was tough going during '39 and '40 but once the USSR was attacked in '41 they folded without too much of a struggle and the game was won - the AAR is up on the forum.
Having said that Alvaro reports he is working on the AI and difficulty levels and things may be more challenging in the future.

[basilstaghare]

My first '39 game as Axis (v1.0) on Historical settings was a walkover and overran much of Russia by late '41 and decided to restart with the latest patch (v1.00.02.01)...Set the AI to Seasoned/Experienced and find it much more challenging...that game is still going...Spring '42 and the Russians are still putting up a good fight...

[elmo3]

The game has been out a while now so how are folks finding the AI? Any improvements since launch? Does it do well on attack and can it make a coherent plan? Thanks.

[AlvaroSousa]

To be clear no game of this complexity can have an A.I. that can beat a human. Not HOI, not SC, not WITE. It is just not possibly currently. There is too much complexity. Even the CIV series which has a vast team of people behind it doesn't have a good military A.I.

I do the best I can with improving the A.I. each go around. So some changes were recently made to how it acts. It is ALWAYS improving. A.I. is just not as efficient as a human player.

If you are an amazing player you will always beat the A.I. There is even a story I read of a game that had an impossible level. The developer said no one would beat the A.I. at this level. Someone did. Even here I thought it would be very hard to beat the A.I. at Ubermensch level with Optimal supply and tyronec did.

[elmo3]

Thanks Alvaro. I'm not looking for Deep Blue (chess computer) level play. Just hoping the AI does not make too many obvious blunders. Chris Park said if anyone beats his game (AI War 2) on level 10 he will consider it a bug and fix it so it doesn't happen again! [:)]

[AlvaroSousa]

I would love a neural net Alpha Go computer to learn WarPlan and play it.

[elmo3]

Well if you can write a deep learning program to play WarPlan I think you could get rich! [;)]

[AlvaroSousa]

If I could write a deep neural net I would be working at Google getting rich.